The Embedded Multi-Core Conference 2020 goes virtual

With the new AUTOSAR Organizational Structure the efforts of various Teams have been joint together and formed the Workgroup Resources (WG-RES). WG-RES now comprises the sub-teams Timing Analysis & Design (TAD), Timing Extensions (TIMEX), AUTOSAR Run-Time Interface (ARTI) as well as AUTOSAR Adaptive Log & Trace (LT). This joint expertise is currently being utilized for describing a unified approach for the Design, Analysis and Trace (i.e. Measurement) of Timing on AUTOSAR Classic as well as Adaptive Platforms. These efforts are concentrated within the AUTOSAR Concept CONC_655 "Unified AUTOSAR Timing and Tracing Approach". The lecture "Unified AUTOSAR Timing and Tracing Approach" explains this Unified Approach of CONC_655 by means of a real Example, including Specification, Tracing and Analysis of a SOME/IP Communication between an AUTOSAR Classic and Adaptive Platform.

 

The automatization of development steps from source-code commit till product delivery, has become a key technology for agile software driven organizations. In automotive systems, safety criticality and mass production scenario results in consideration of new challenges at continuous software development of AUTOSAR systems. In this talk, we show how AUTOSAR Application Software integration can be shifted in major parts from manual integration to continuous integration and how further development steps, like non-functional timing tests can be migrated to the continuous approach.

 

New trends with Domain and Zone architecture or general ECU consolidation create new demands for more complex microcontrollers and software strategies. ECU virtualization is one of them. Hardware support for virtualization decreases the software overhead and creates system more safe with strict software separation. Microcontrollers are in front of another era than "just" a multicore and higher frequency.Virtualization in the world of automotive MCUs.

 

Heterogeneous computing has become a mainstay of personal computers with support from CPU-GPU architectures. While the main applications run on multi-core CPUs, computationally intensive tasks demonstrating data level parallelism are typically offloaded to GPUs. There are several software frameworks available today to support the aforementioned style of computation. Examples include OpenMP, OpenCL, OpenAMP, CUDA etc. We introduce OpenCL and discuss its adoption for both heterogeneous and homogeneous computing on embedded microcontrollers. With this, a software architect can derive the maximum bang for the buck by a careful concurrent spread of execution of application functional blocks on available and future compute units.OpenCL for heterogeneous computing on embedded controllers.